Think with 3D: Geometric Imagination Grounded Spatial Reasoning from Limited Views

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Chen, Zhangquan, Zhang, Manyuan, Yu, Xinlei, Luo, Xufang, Sun, Mingze, Pan, Zihao, An, Xiang, Feng, Yan, Pei, Peng, Cai, Xunliang, Huang, Ruqi
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866914389435613184
author Chen, Zhangquan
Zhang, Manyuan
Yu, Xinlei
Luo, Xufang
Sun, Mingze
Pan, Zihao
An, Xiang
Feng, Yan
Pei, Peng
Cai, Xunliang
Huang, Ruqi
author_facet Chen, Zhangquan
Zhang, Manyuan
Yu, Xinlei
Luo, Xufang
Sun, Mingze
Pan, Zihao
An, Xiang
Feng, Yan
Pei, Peng
Cai, Xunliang
Huang, Ruqi
contents Though recent advances in vision-language models (VLMs) have achieved remarkable progress across a wide range of multimodal tasks, understanding 3D spatial relationships from limited views remains a significant challenge. Previous reasoning methods typically rely on pure text (e.g., topological cognitive maps) or on 2D visual cues. However, their limited representational capacity hinders performance in specific tasks that require 3D spatial imagination. To address this limitation, we propose 3DThinker, a framework that can effectively exploits the rich geometric information embedded within images while reasoning, like humans do. Our framework is the first to enable 3D mentaling during reasoning without any 3D prior input, and it does not rely on explicitly labeled 3D data for training. Specifically, our training consists of two stages. First, we perform supervised training to align the 3D latent generated by VLM while reasoning with that of a 3D foundation model (e.g., VGGT). Then, we optimize the entire reasoning trajectory solely based on outcome signals, thereby refining the underlying 3D mentaling. Extensive experiments across multiple benchmarks show that 3DThinker consistently outperforms strong baselines and offers a new perspective toward unifying 3D representations into multimodal reasoning. Our code is available at https://github.com/zhangquanchen/3DThinker.
format Preprint
id arxiv_https___arxiv_org_abs_2510_18632
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Think with 3D: Geometric Imagination Grounded Spatial Reasoning from Limited Views
Chen, Zhangquan
Zhang, Manyuan
Yu, Xinlei
Luo, Xufang
Sun, Mingze
Pan, Zihao
An, Xiang
Feng, Yan
Pei, Peng
Cai, Xunliang
Huang, Ruqi
Computer Vision and Pattern Recognition
Artificial Intelligence
I.2.10
Though recent advances in vision-language models (VLMs) have achieved remarkable progress across a wide range of multimodal tasks, understanding 3D spatial relationships from limited views remains a significant challenge. Previous reasoning methods typically rely on pure text (e.g., topological cognitive maps) or on 2D visual cues. However, their limited representational capacity hinders performance in specific tasks that require 3D spatial imagination. To address this limitation, we propose 3DThinker, a framework that can effectively exploits the rich geometric information embedded within images while reasoning, like humans do. Our framework is the first to enable 3D mentaling during reasoning without any 3D prior input, and it does not rely on explicitly labeled 3D data for training. Specifically, our training consists of two stages. First, we perform supervised training to align the 3D latent generated by VLM while reasoning with that of a 3D foundation model (e.g., VGGT). Then, we optimize the entire reasoning trajectory solely based on outcome signals, thereby refining the underlying 3D mentaling. Extensive experiments across multiple benchmarks show that 3DThinker consistently outperforms strong baselines and offers a new perspective toward unifying 3D representations into multimodal reasoning. Our code is available at https://github.com/zhangquanchen/3DThinker.
title Think with 3D: Geometric Imagination Grounded Spatial Reasoning from Limited Views
topic Computer Vision and Pattern Recognition
Artificial Intelligence
I.2.10
url https://arxiv.org/abs/2510.18632